Andrew Davison

Andrew Davison

@ajdavison.bsky.social

From SLAM to Spatial AI; Professor of Robot Vision, Imperial College London; Director of the Dyson Robotics Lab; Co-Founder of Slamcore. FREng, FRS.

1.8K followers1.4K following57 posts

Andrew Davison
@ajdavison.bsky.social

My first computer programming was Dragon 32 Basic, back in about 1982!

The Centre for Computing History@computinghistory.org.uk

They're back! We're very happy to be hosting this year's Dragon Meetup. You don't have to be a Dragon owner - or even know much about Dragons - to come to this celebration of the classic Welsh home computer. www.computinghistory.org.uk/det/76704/dr...

Andrew Davison
@ajdavison.bsky.social

I've been a Tamiya RC fan since owning and loving a Grasshopper in the 80s, and more recently built one or their RC kits with each of my kids (Hornet, LunchBox and Midnight Pumpkin!)

Matt Alt@mattalt.bsky.social

RIP Shunsaku Tamiya, age 90, the chairman of legendary modelmaker Tamiya. He didn't found the company, but started working there in 1958, just four years after its founding, and helped it shift from its original line of wooden models into the plastic kits it is so well known for today.

Andrew Davison
@ajdavison.bsky.social

Live Demo of MASt3R-SLAM this afternoon in the demo session at #CVPR2025.

Eric Dexheimer@ericdexheimer.bsky.social

We’ve had fun testing the limits of MASt3R-SLAM on in-the-wild videos. Here’s the drone video of a Minnesota bowling alley that we’ve always wanted to reconstruct! Different scene scales, dynamic objects, specular surfaces, and fast motion.

Andrew Davison
@ajdavison.bsky.social

Full draft of the SLAM Handbook now released --- available as a free PDF, with a printed version coming soon. Now including Part 3, "From SLAM to Spatial AI" (I knew it would catch on eventually), with contributions from @hidematsu82.bsky.social and me. #SpatialAI github.com/SLAM-Handboo...

github.comGitHub - SLAM-Handbook-contributors/slam-handbook-public-release: Release repo for our SLAM HandbookRelease repo for our SLAM Handbook. Contribute to SLAM-Handbook-contributors/slam-handbook-public-release development by creating an account on GitHub.
Andrew Davison
@ajdavison.bsky.social

Great diagram.

Brent Toderian@brenttoderian.bsky.social

This is interesting — a recent study of mode share (the % share of transportation trips by car, transit, walking, biking etc) relative to city size and income levels in almost 800 cities in 61 countries. Interesting results. HT @davidzipper.bsky.social www.sciencedirect.com/science/arti...

Andrew Davison
@ajdavison.bsky.social

Great look at back at 2D scrolling methods in games.

Itay Keren@untame.com

Hard to believe that a decade ago, in preparation for the release of my game, I spent a month researching 2D cameras by playing through all my favorite 80s classics. If you haven't yet, check out this fun, GIF-heavy paper! docs.google.com/document/d/e...

Andrew Davison
@ajdavison.bsky.social

We were fortunate to have Jianyuan visit us at Imperial yesterday, and he gave a brilliant and thoughtful talk about 3D geometry and learning. Maybe every 6--12 months or so there is a paper in my area which really seems like something new and important, and VGGT is one of those!

Jianyuan Wang@jianyuanwang.bsky.social

Introducing VGGT (CVPR'25), a feedforward Transformer that directly infers all key 3D attributes from one, a few, or hundreds of images, in seconds! Project Page: vgg-t.github.io Code & Weights: github.com/facebookrese...

Andrew Davison
@ajdavison.bsky.social

Should be good!

Matthias Niessner@niessner.bsky.social

Tomorrow in our TUM AI - Lecture Series we'll have Andrea Tagliasacchi (@taiyasaki.bsky.social), SFU. He'll talk about "𝐑𝐚𝐝𝐢𝐚𝐧𝐭 𝐅𝐨𝐚𝐦: 𝐑𝐞𝐚𝐥-𝐓𝐢𝐦𝐞 𝐃𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭𝐢𝐚𝐛𝐥𝐞 𝐑𝐚𝐲 𝐓𝐫𝐚𝐜𝐢𝐧𝐠". Live Stream: youtube.com/live/1u7ahb9... 5pm GMT+1 / 9am PST (Mon Mar 24th)

youtube.comTUM AI Lecture Series - Radiant Foam: Real-Time Differentiable Ray Tracing (Andrea Tagliasacchi)YouTube video by Matthias Niessner
Andrew Davison
@ajdavison.bsky.social

Remember: the route to Spatial AI is through SLAM. Key research topics are representations for real-time fusion of measurements and priors into persistent semantic/metric scene models; and algorithm/hardware co-design for 1000X efficiency. #SpatialAI arxiv.org/abs/1803.11288

arxiv.orgFutureMapping: The Computational Structure of Spatial AI SystemsWe discuss and predict the evolution of Simultaneous Localisation and Mapping (SLAM) into a general geometric and semantic `Spatial AI' perception capability for intelligent embodied devices. A big ga...

Andrew Davison reposted

Chris Offner
@chrisoffner3d.bsky.social

Replying to @chrisoffner3d.bsky.social

On the tram station video, VGGT fails entirely, whereas MASt3R-SLAM had done a more reasonable (albeit also not stellar) job. bsky.app/profile/chri...

Chris Offner@chrisoffner3d.bsky.social

This underground tram station also shows some drifting, and some scene elements float in places where they don't belong. The overall layout is correct though, unlike in the (much longer) video of the ETH main building that I showed first.

Andrew Davison
@ajdavison.bsky.social

I remember seeing this drone video a few years ago and thinking "we'll never run SLAM on that".... but here it is, complete with dense reconstruction (single camera, unknown calibration, no IMU). MASt3R-SLAM is absurdly robust.

Eric Dexheimer@ericdexheimer.bsky.social

We’ve had fun testing the limits of MASt3R-SLAM on in-the-wild videos. Here’s the drone video of a Minnesota bowling alley that we’ve always wanted to reconstruct! Different scene scales, dynamic objects, specular surfaces, and fast motion.

Andrew Davison
@ajdavison.bsky.social

Open source code now available MASt3R-SLAM: the best dense visual SLAM system I've ever seen. Real-time and monocular, and easy to run with a live camera or on videos without needing to know the camera calibration. Brilliant work from Eric and Riku.

Riku Murai@rmurai0610.bsky.social

MASt3R-SLAM code release! github.com/rmurai0610/M... Try it out on videos or with a live camera Work with @ericdexheimer.bsky.social*, @ajdavison.bsky.social (*Equal Contribution)

Andrew Davison
@ajdavison.bsky.social

Wow I'd like to try to understand this. Something for @tinytapeout.com ?

Sung Kim@sungkim.bsky.social

tiny-gpu A minimal GPU in Verilog optimized for learning about how GPUs work from the ground up. Built with <15 files of fully documented Verilog, complete documentation on architecture & ISA, working matrix addition/multiplication kernels, and full support for kernel simulation & execution traces

Andrew Davison
@ajdavison.bsky.social

Looks like a great workshop, happening in Malaga this July.

Sebastian Risi@risi.bsky.social

We're excited to announce the first Evolving Self-organisation workshop at GECCO 2025! Submission deadline: March 26, 2025 More information: evolving-self-organisation-workshop.github.io

Andrew Davison
@ajdavison.bsky.social

Yes, software for robotics (and vision) comes and goes and in the end you'll have to build it yourself if you want to really push forwards.

Rodney Brooks@rodneyabrooks.bsky.social

1/5 I love NVIDIA hardware. At Robust.AI we use it on-robot for vision, neural networks, & planning. Someone's marketing team beats me up on why we don't use their robot programming system. I have replied "it doesn't get me excited". "But why not?" they incredulously say. I now have an answer.

Andrew Davison
@ajdavison.bsky.social

I'm recruiting a new PhD student on Gaussian Belief Propagation for Spatial AI. For this particular opportunity: 1. The funding I have is only available to a student who qualifies for "home fees" in England; 2. I'm looking for someone who can start soon, ideally April 2025.

Andrew Davison
@ajdavison.bsky.social

I'm looking for someone with a strong academic and technical background (e.g. computing, maths, physics or engineering) but not necessarily specific prior work in AI. Please DM or email if you are interested.

Andrew Davison
@ajdavison.bsky.social

Nice analogy. I'm wondering what the future of scientific publishing should be, given that LLMs are increasingly good at automatically going from full paper to summary, or vice versa. Surely we could be publishing new ideas in the "sketch" form only, but I can't quite imagine exactly what that is.

Peyman Milanfar@docmilanfar.bsky.social

Picasso's le taureau study is a masterful illustration of distilling an idea to its core essence. A deliberate and careful stripping away of details - especially when exposing an idea in a paper or lecture - can lead to clarity, a more powerful message, and impactful insights

Andrew Davison
@ajdavison.bsky.social

I was surprised that on Strava my best average speed for 10km was a bit faster than for 5 miles (which is shorter, about 8km). Strava automatically finds the fastest segments from any ride you've recorded. Normally of course you see a faster average speed for shorter distances. 1/n

Andrew Davison
@ajdavison.bsky.social

But I was confused about if it's even possible to have a faster average speed for a longer distance. When the distance changes by a factor of 2, it's not. Take your fastest 10km segment, split it into two 5km segments, and either both will have the same average or one will be faster, one slower. 2/n

Andrew Davison
@ajdavison.bsky.social

However, if the factor of distance difference is less than 2, like 10 and 8, you can indeed have a faster average for 10km. Consider a 10km ride with very fast first and last km but slow in between. Would be a weird way to ride... 3/3

Andrew Davison reposted

Melanie Mitchell
@melaniemitchell.bsky.social

Excellent post about the recent OpenAI o3 results on ARC (& other benchmarks). I don't know how @natolambert.bsky.social manages to write these so quickly! I highly recommend his newsletter. www.interconnects.ai/p/openais-o3... I am (more slowly) writing my own take on all this, coming soon.

interconnects.aio3: The grand finale of AI in 2024A step change as influential as the release of GPT-4. Reasoning language models are the current big thing.
Andrew Davison
@ajdavison.bsky.social

One of the landmark games of all time for me that I spent hours playing on my friend's BBC micro. 3D vector graphics, trading, upgrades and thousands of procedurally generated planets all in 32k, incredible.

proc_gen@procgen.bsky.social

Mark Moxon has been throughly documenting the source code of the various versions of the original Elite. elite.bbcelite.com I think at this point the game may be one of the most throughly documented games of all time.

elite.bbcelite.comFully documented source code for Elite on the 6502 (BBC Micro, Acorn Electron, Commodore 64, NES)Fully documented source code for Elite on the 6502 (BBC Micro, Acorn Electron, Commodore 64, NES), with more than 100 deep dive articles explaining how every aspect of this iconic game works.
Andrew Davison
@ajdavison.bsky.social

MASt3R-SLAM: easily the most robust dense monocular SLAM system I've ever seen, and also very accurate. Real-time, RGB only (no IMU), handles unknown focal length and zoom. More new videos coming soon @ericdexheimer.bsky.social @rmurai0610.bsky.social. Paper and more: edexheim.github.io/mast3r-slam

edexheim.github.ioMASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction PriorsMASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors
Riku Murai@rmurai0610.bsky.social

Introducing MASt3R-SLAM, the first real-time monocular dense SLAM with MASt3R as a foundation. Easy to use like DUSt3R/MASt3R, from an uncalibrated RGB video it recovers accurate, globally consistent poses & a dense map. With @ericdexheimer.bsky.social* @ajdavison.bsky.social (*Equal Contribution)

Andrew Davison
@ajdavison.bsky.social

I'd love to see all of this talk (public link?) No we are not running out of "data", but human-written text is unique data which has been supervised by intelligence (it's ground truth for general intelligence, or at least part of it), and very different therefore from passively captured video.

Dhruv Batra@dhruvbatra.bsky.social

Brilliant talk by Ilya, but he's wrong on one point. We are NOT running out of data. We are running out of human-written text. We have more videos than we know what to do with. We just haven't solved pre-training in vision. Just go out and sense the world. Data is easy.

Andrew Davison
@ajdavison.bsky.social

Nice explanation of information entropy.

Keenan Crane@keenancrane.bsky.social

Entropy is one of those formulas that many of us learn, swallow whole, and even use regularly without really understanding. (E.g., where does that “log” come from? Are there other possible formulas?) Yet there's an intuitive & almost inevitable way to arrive at this expression.

Andrew Davison
@ajdavison.bsky.social

This looks great!

Maurice Fallon@mauricefallon.bsky.social

To kick off using Bluesky: our new dataset called Oxford Spires. Synchronised, multi-color cameras and lidar in multiple Oxford colleges. Ground-Truth highly accurate 3D maps from tripod scanners. The ideal basis for NeRF/3DGS SLAM research. dynamic.robots.ox.ac.uk/datasets/oxf...